Embodiment
Here exemplary embodiment will be illustrated in detail, its example is illustrated in the accompanying drawings.Following description is related to
During accompanying drawing, unless otherwise indicated, the same numbers in different accompanying drawings represent same or analogous key element.Following exemplary embodiment
Described in embodiment do not represent all embodiments consistent with this specification.On the contrary, they are only and such as institute
The example of the consistent apparatus and method of some aspects be described in detail in attached claims, this specification.
It is only merely for the purpose of description specific embodiment in the term that this specification uses, and is not intended to be limiting this explanation
Book." one kind " of used singulative, " described " and "the" are also intended to bag in this specification and in the appended claims
Most forms are included, unless context clearly shows that other implications.It is also understood that term "and/or" used herein is
Refer to and any or all may be combined comprising the associated list items purpose of one or more.
It will be appreciated that though various information may be described using term first, second, third, etc. in this specification, but
These information should not necessarily be limited by these terms.These terms are only used for same type of information being distinguished from each other out.For example, do not taking off
In the case of this specification scope, the first information can also be referred to as the second information, and similarly, the second information can also be claimed
For the first information.Depending on linguistic context, word as used in this " if " can be construed to " ... when " or
" when ... " or " in response to determining ".
The contents such as the page and menu item of more, each function correlations of function that the APP of many user side equipment installings has
It is more, and the characteristic that nest relation between the page, menu is more complicated.In view of this, user very may be used during using APP
The problem of more can be run into, such as:It can not find the application contents such as the page related to some functions, menu item.In order to solve
The problem of running into, user, which can typically make repeated attempts, clicks on the APP various pages and menu, and after attempting to have no result, user may ask
Help customer service, by describing the problem of oneself runs into accordingly to be replied, or the information related to problem is input to and searched
The rope page, then browses the various search results of searched page presentation, then selects the search knot that can solve its problem
Fruit.This process for solving problem can need to expend longer time, and need user to perform complicated operation.
In order to avoid user's consuming longer time, complicated operation is performed, this specification embodiment, Fig. 1 can be passed through
Next time behavior of the server prediction user at current time in shown application scenarios, the behavior can reflect that user is current
The intention at moment, for solving the problems, such as that user runs into.
In order to predict user behavior, the user that server can upload beforehand through the APP installed in each user side equipment
Behavior sequence, train forecast model, after training forecast model, the model of training is issued to user side equipment, by user
Side apparatus with reference to the behavior sequence after user, predicts the behavior next time at current time according to the forecast model trained, and
There is provided for realizing the quick operating mode of behavior next time predicted, the user without user side equipment seek help customer service or
Input the problem of being run into user related information, you can quickly solve its problem, make repeated attempts click APP's without user
The various pages and menu.
In practical application, as shown in figure 1, being equiped with APP1 user side equipment 1 and being equiped with APP2 user side equipment 2
Respectively by step J311, J321, the behavioural information of user is recorded to local user's user behaviors log, the behavior for generating user is remembered
Record, Tables 1 and 2 as shown in Figure 1, behavioural information can be that behavior corresponding to the behavior at user's each moment identifies, behavior
Mark in chronological sequence sort, the behavior sequence of composition, as illustrated in the drawing A1A2A3A4A5......,
B1B2B3B4B5......, in some examples, behavior mark corresponding to behavior can be according to specific APP types and application
The bibliography system of scene, set corresponding to each class behavior of user behavior to identify, by APP developer when burying daily record just
Pre-define.Such as:In Alipay APP, 10000 represent and transfer accounts behavior, in other examples, can also be made by other symbols
Identified for user, this specification embodiment is without limitation.
After the certain period for generating user behavior record, (user touches in real time or when meeting that predetermined behavior uploads condition
Send out upload button or reach predetermined uplink time, it may be determined that meet that predetermined behavior record uploads condition), user side equipment 1
, can be respectively from the row of local log read user with user side equipment 2 respectively by step J112 and J113, J122 and J123
For information, then sort in chronological order, constituting action sequence, and upload onto the server respectively, then server performs step
J101, based on the forecast model trained, the demand of user 1 and user 2 is predicted respectively with reference to the behavior sequence that rigid connection is received, then divide
The demand of prediction is not issued to by user side equipment 1 and user side equipment 2 by step J114 and J124.
In another example, the behavioural information of user side equipment record user is to local user's user behaviors log, except predetermined
Behavior mark is outer, and the behavior record of the user of generation can also be other character informations of the behavior at description user's each moment,
When meeting that predetermined behavior uploads condition, user side equipment can be according to the record time by evening to early order, from being locally stored
User action log, read current time before at least two behavior records;According between user behavior and behavior mark
Predetermined corresponding relation, it is determined that each read records behavior mark corresponding to recorded historical behavior;By the behavior of determination
Mark sorts according to reading order, forms the behavior sequence of the user.Then uploaded to server.
In other examples, user side equipment directly can also upload onto the server the behavior record of user, by servicing
Device is according to the predetermined corresponding relation between historical behavior and behavior mark, it is determined that each read records recorded historical behavior
Corresponding behavior mark;The behavior of determination is identified and sorted according to reading order, forms the behavior sequence of the user.
In summary, after forecast model is issued to user side equipment by the server in the application scenarios shown in Fig. 1, this theory
Scheme disclosed in bright book directly can obtain the behavior sequence of user in user side;Then by being stored in local prediction mould
Type, user is predicted in the behavior next time at current time, further according to prediction result, there is provided predicted down for realizing
The quick operating mode of behavior, complicated operation is carried out without user, can be quick by the quick operating mode of offer
Completing user needs operation to be performed.The process of the offer user prompt operation of this specification is provided in detail below in conjunction with accompanying drawing.
Fig. 2A is referred to, Fig. 2A is the method for the offer user prompt operation shown in the exemplary embodiment of this specification one
Flow chart, this method are applied to user side equipment, may comprise steps of S201-S203:
Step S201, the behavior sequence of user is obtained, wherein, the behavior sequence includes the behavior letter before current time
Breath.
Step S202, behavior sequence input is stored in local forecast model, to user under current time
Behavior is predicted.
Step S203, according to prediction result, there is provided for realizing the quick operating mode for the behavior next time predicted.
This specification embodiment, the equipment for being equiped with default application program is can apply to, other can also be installed on and set
Standby, the other equipment can be by data wire or other connected modes not influenceed by network traffics, with being equiped with default answer
Connected with the equipment of program.Here default application program can be the application program that the operating system of equipment carries, such as:Intelligence
The voice assistant of mobile phone or the various application programs of third party's exploitation.
The application program carried for the operating system of equipment, when recording the behavioural information of user, it can detect in real time simultaneously
Record behavioural information of the user to operation interface in itself, referred to herein as behavioural information can refer to description user to operation interface
The character information of the operation such as slip, mobile, addition, deletion, switching, when recording the behavioural information of user, can also be detected simultaneously in real time
User is recorded to the behavioural information of control contained by operation interface, referred to herein as behavioural information can refer to description user to operating boundary
The character information of the operations such as the striking of control contained by face, deletion;When recording the behavioural information of user, it can also detect and remember in real time
Employ behavioural information of the family to the shortcut of equipment.Shortcut such as home keys, return key, tuning key etc..
For the various application programs (APP) of third party's exploitation, when recording the behavioural information of user, can detect in real time simultaneously
The overall behavioural information in each interface of the user to APP is recorded, can also in real time detect and record user to control contained by APP each interface
The behavioural information of part.Such as:User is in behavioural informations such as the browsing of APP interface, clicks to control.
After user side carries out the behavior record of user, it is contemplated that user behavior can reflect user view, for ease of and
When predict user view, it is predetermined before current time that this specification embodiment can read user from User action log
The behavioural information of item number, referred to herein as behavioural information can be can characterize the historical behavior of user behavior identify, also may be used
Be description description user historical behavior other character informations.The item number of the behavioural information read can be by this programme
Designer is set according to specific application scenarios, when such as predetermined application being voice customer service, the number of the behavioural information of reading
Mesh could be arranged to 6, in other examples, the number of the behavioural information of reading can also be arranged into other numerical value, this explanation
Book embodiment is without limitation.
After reading the behavioural information of user, in order to the behavioural information based on user obtain can effecting reaction user it is current when
The behavior next time of the intention at quarter, can be converted to behavior sequence by the behavioural information of reading and be input to and be stored in local prediction
Model, user is predicted in the behavior next time at current time.Referred to herein as forecast model can be reality that Fig. 1 is related to
The forecast model mentioned in example is applied, institute is trained according to the behavioural information in the User action log of user side equipment by server
.In other examples, gained, this theory can also be trained by the other equipment in addition to server by being stored in local forecast model
Bright book embodiment is without limitation.
Server can train the wider forecast model of an applicability when training forecast model, and the forecast model is fitted
For a variety of users, a variety of application programs, in addition, it is contemplated that the memory data output and/or data processing speed of user side equipment
Less than server, the forecast model matched with the memory data output of user side equipment and/or data processing speed can also be trained,
Accelerate the behavior prediction speed of forecast model.
In order to train the forecast model that can accelerate behavior prediction speed, server can be to suitable for a variety of users, more
The forecast model of kind of application program is compressed, and during actual compression institute, can be deleted small feature is influenceed on modelling effect, or
Person, which retains, influences big top n feature on modelling effect, and the capacity of model is compressed in certain limit.
Further, it is also possible to for the actual conditions of different types of user side equipment, predetermined speed can be accelerated by training,
The forecast model matched respectively with different types of user side equipment again.The type of user side equipment by user side equipment application
Type (APP type) and/or user characteristics decision, such as:The equipment for being provided with data search application sets for a kind of user side
Standby, the equipment for being provided with navigation application is another kind of user side equipment;For another example:Data search application, user 30 are installed
It is a kind of user side equipment to the equipment of 40 years old, data search application is installed, the equipment that user is 20 to 30 years old is another kind of
User side equipment.
The feature included in the forecast model matched with certain class user side equipment, can be by the application of such user side equipment
At least one in the User action log of type, the user characteristics of such user side equipment and such user side equipment determines.
Wherein, the application type of user side equipment can refer to the type of the mounted application program of user side equipment, can also refer to and wait to pacify
The type of the application program of dress;User characteristics can include the age of user, surname not, educational background, occupation, income, custom, partially
The feature such as good.
In some examples, train matched with certain class user side equipment forecast model the step of can be by as shown in Figure 1
Server perform, the forecast model after training is issued to user side equipment by the server again, describes forecast model below
Training step:
Obtain the behavior sequence of the user of fellow users side apparatus, N+1 items behavioural information contained by the sequence temporally from
The early order sequence to evening, N is the natural number more than 1.
The behavior sequence gathered is divided into training sample and test sample.
By preceding N items behavioural information in the behavior sequence of each user in the training sample, distinguish as training sequence defeated
Enter predetermined forecast model to be trained.
Based on N+1 items behavioural information in the behavior sequence of each user in the training sample and training forecast model institute
The ProbabilityDistribution Vector obtained, the ProbabilityDistribution Vector includes the probability that the training sequence is mapped to each behavior, after obtaining training
Forecast model.
Test sample is concentrated to preceding N items behavioural information in the behavior sequence of each user, instruction is input to as cycle tests
The forecast model (model parameter after training) perfected, obtains the ProbabilityDistribution Vector of forecast model output, the probability distribution
Vector is mapped to the probability of each behavior comprising the cycle tests.
N+ in the behavior sequence of each user is concentrated based on ProbabilityDistribution Vector, test sample obtained by test forecast model
1 behavioural information, the predictablity rate of valuation prediction models.
And then the modes such as the additions and deletions aspect of model, the positive and negative sample proportion of adjustment, grid search model hyper parameter can also be utilized
Above-mentioned training process is repeated, filters out and optimal model is showed on test set.
Further, it is also possible to the predictablity rate of forecast model is weighed using model-evaluation indexes such as AUC, F1SCORE.
Wherein, can be by installing in user side equipment when obtaining the behavior sequence of user of fellow users side apparatus
Application records and the behavioural information for transmitting user, in other embodiment, other modes can also be used to obtain behavior sequence,
This specification is without limitation.
Above-mentioned forecast model is to temporally the behavior sequence of sequence gained is handled sooner or later, in order to improve processing effect
Rate, the model for comparing and being good at the behavior sequence that processing according to time sequence generates can be chosen, for example, can choose following any
Model or at least two models are combined as the forecast model:
Shot and long term memory models, gradient lifting decision tree, Logic Regression Models.
When forecast model is shot and long term memory models LSTM models (shot and long term memory models), model may include an input
Layer, an embedding+dropout layers, a lstm+dropout layers, a prediction interval connected entirely, a softmax loss layers and
One accuracy computation layers.When forecast model is shot and long term memory models LSTM, by compact model, the shot and long term note trained
30M can be less than by recalling model LSTM capacity.
In other examples, the step of training forecast model, can be performed by the other equipment outside server, and training is pre-
The mode of model is surveyed, the other technologies means of this area can also be taken., can commenting according to model when training forecast model
Valency index determines the probability threshold value of model.Such actual prediction user is in the behavior next time at current time, if user
When the probability that behavior sequence is mapped to certain behavior is more than the threshold value, then judge that the behavior is going next time for user's current time
For.If the probability for being mapped to multiple behaviors is all higher than threshold value, it is determined that when that behavior of maximum probability is that user is current
The behavior next time carved.In other examples, based on the ProbabilityDistribution Vector, probability can also be expired by other means
The behavior of the predetermined predicted condition of foot is defined as next time behavior of the user at current time, such as:By the behavior of maximum probability
It is determined that into the behavior for meeting predetermined predicted condition.
After training forecast model, the forecast model trained can be issued to user side equipment by server, by user
Its local is arrived in side apparatus storage, when needing to predict user behavior, the behavior sequence of user is directly obtained, by being stored in local
Forecast model carry out behavior prediction, prediction result can be the behavior that predicted value is more than threshold value, and in some examples, predicted value can
To be probability that the behavior sequence of user is mapped to user behavior, threshold value can be probability threshold value.
In practical application, during Descend Prediction model, server can apply journey receiving user side equipment request installation
During sequence, the application program packing of the forecast model and request installation is issued to the user side equipment and is dealt into the user side
Equipment.Forecast model individually can also be issued to user side equipment.In addition, server is detecting the forecast model by more
After new, the forecast model after renewal can also be issued to the user side equipment.
If server, for the actual conditions of different types of user side equipment, trains when training forecast model
The forecast model matched with certain class user side equipment, can by each forecast model trained respectively with all types of user side apparatus
At least one corresponding storage in application type, user characteristics, User action log, can be according to be received pre- when issuing
At least one in the application type, user characteristics and User action log of the user side equipment of model is surveyed, from various prediction moulds
Forecast model to be issued is selected in type.
After forecast model is issued into user side equipment, user side equipment can be at once or in predetermined behavior prediction
Time or response user's triggering, start the program of prediction user behavior using forecast model.It is pre- starting in some examples
Survey user behavior program after, can according to record the time from the User action log being locally stored, read current time it
The behavior record of preceding predetermined item number;According to the predetermined corresponding relation between user behavior and behavior mark, it is determined that is read is every
Item records behavior mark corresponding to recorded historical behavior;By the behavior mark of determination according to record time-sequencing, forming should
The behavior sequence of user.Wherein, can be read during reading by the record time by the order in evening to morning.In other examples, also may be used
Sequentially read with order between being late for, sorted in chronological order after generation behavior mark, time series can also be obtained.
Some examples, the forecast model that user side equipment is stored can be the LSTM models trained, it is assumed that Yong Hu
t0, t1, t2..., tn-1The behavior at moment identifiesThe mark at these moment is temporally arranged sooner or later
After sequence, form the behavior sequence of user, the LSTM models that the sequence inputting trains calculated, obtain probability distribution to
Amount, the ProbabilityDistribution Vector include the probability that the behavior sequence is mapped to each behavior.
Further, since the behavior next time at user's current time can reflect the problem of user is likely encountered at current time,
In some examples, this programme can be using ProbabilityDistribution Vector as each behavior predicted, reflection user tnWhat the moment was likely encountered
Each problem needs the score value answered, and calculation formula can be as follows:
If user is in tnMoment generates behavior againThen can to newest n, (n be more than 2 by forecast model
Natural number) behavior sequence of individual behavior mark composition calculated, obtain reflection user tn+1Each problem that moment is likely encountered needs
The score value to be answered:
Forecast model output probability distribution vector, this programme can be based on the ProbabilityDistribution Vector, probability be met pre-
The behavior of fixed predicted condition is defined as next time behavior of the user at current time.
During the behavior next time at actual determination user's current time, difference can be used really according to different predicted conditions
Determine mode, in some examples, the ProbabilityDistribution Vector can be based on by following operation, probability is met to predetermined prediction bar
The behavior of part is defined as next time behavior of the user at current time:
Compare the magnitude relationship that the behavior sequence is mapped to the probability and predetermined probability threshold value of each behavior.
If the probability that the behavior sequence is mapped to any behavior is more than the probability threshold value, it is determined that any row
To meet predetermined predicted condition.
Wherein, predetermined predicted condition and probability threshold value, can be by the designer of this programme according to actual application
Scene settings, such as:Probability threshold value is set as 80%.
After the behavior next time for determining user's current time, in view of the behavior next time predicted can reflect that user is current
The demand at moment or the problem of run into, can be according to prediction result, there is provided for realizing the quick of the behavior next time predicted
Mode of operation, and then the needs of user's current time is met based on the quick operating mode or solves the problems, such as that user runs into.
In some examples, according to prediction result next time, there is provided during quick operating mode, user can be saved and gone
For the operation of required execution, the mode that can quickly perform order corresponding to the row next time predicted directly is provided, such as:Display
The button of order, can also directly initiate the behavior pair next time predicted corresponding to the behavior next time that quick response is predicted
The mode for the software program answered, such as:Jump directly to the page corresponding to the behavior next time predicted.
In addition, when providing quick operating mode, the word of the behavior next time for describing to be predicted can also be first exported
After according with information and representing that the button of prediction correctness, user confirm, then directly initiate the behavior next time predicted and correspond to
Software mode., can be by following operation according to prediction result, there is provided predicted for realization next in some examples
The quick operating mode of secondary behavior:
It is determined that the search key related to the behavior next time predicted.
Identified search key is added to search box.
Wherein it is possible to the corresponding relation between various actions and its search key is preset, such as:By various actions phase
The keyword of pass is arranged to order corresponding with the row, or software program corresponding with the behavior.
In some scenes, it can be exported the related keyword of the behavior predicted as the placeholder of search input frame,
Fig. 2 B specifically are see, APP shown in figure is Alipay, and " shared bicycle " is searched in the behavior predicted, " shared bicycle " is made
It is presented on for placeholder in search input frame.Prompt user's search shared bicycle., can be with pop-up, voice in other examples
The search key related to the behavior next time predicted is exported etc. mode, this specification embodiment is without limitation.
The process of the offer user prompt operation of this specification embodiment is provided below in conjunction with the application scenarios shown in Fig. 3.
The server in application scenarios shown in Fig. 3, the user that the APP of installation in each user side equipment is uploaded can be received
Behavior sequence, then perform step J300 training forecast model, after training forecast model, can perform step J311 and
J321, the forecast model trained is issued to user side equipment 1 and user side equipment 1 respectively, by user side equipment 1 and user
Side apparatus 2 performs step J312 and J322 respectively, stores the forecast model trained to locally, is then distinguished by APP1 and APP2
Step J313 to step J315, step J323 to step J323 to step J325 are performed, from the user behavior of local daily record storage
In record, temporally order reads at least two behavioural informations before current time sooner or later, forms the behavior sequence of user, so
The forecast model that behavior sequence input is locally stored afterwards, predicts the behavior next time at user's current time, then provide and be used for
The quick operating mode of behavior next time, the user without user side equipment seek help from customer service or will be with problem phases described in realizing
The information of pass is input to searched page, you can quickly solves user's request, is made repeated attempts without user and click on APP various pages
Face and menu.
In addition, user side equipment 1 and user side equipment 2 can also perform step J316 and J326 respectively, by local daily record
The user behavior information newly stored is uploaded onto the server, and is continued to predict mould according to the user behavior information updating of upload by server
Type, and the forecast model of renewal is issued to user side by request server.
Corresponding with the embodiment of preceding method, this specification additionally provides the embodiment of device.
Referring to Fig. 4, Fig. 4 is the logic of the device of the offer user prompt operation shown in the exemplary embodiment of this specification one
Block diagram, the device 400 are applied to user side equipment, can included:Retrieval module 410, behavior prediction module 420 and operation
Module 430 is provided.
Wherein, retrieval module 410, for obtaining the behavior sequence of user, wherein, the behavior sequence includes current
Behavioural information before moment.
Behavior prediction module 420, for behavior sequence input to be stored in into local forecast model, user is being worked as
The behavior next time at preceding moment is predicted.
Operation provides module 430, for according to prediction result, there is provided for realizing the quick of the behavior next time predicted
Mode of operation.
In some examples, operation, which provides module 430, to be included:
Keyword determining module, for determining the search key related to the behavior next time predicted;
MIM message input module is searched for, for identified search key to be added into search box.
In other examples, retrieval module 410 can include:
Behavior record read module, for from the User action log being locally stored, being read current according to the record time
The behavior record of predetermined item number before moment;
Behavior identifies determining module, for the predetermined corresponding relation between being identified according to user behavior and behavior, it is determined that being read
Each taken records behavior mark corresponding to recorded user behavior;
Sequence composition module, for according to time-sequencing is recorded, the behavior mark of determination to be formed into the behavior sequence of the user
Row.
In other examples, behavior prediction module 420 can include:
Sequence computing module, calculated for behavior sequence input to be stored in into local forecast model;
Vectorial acquisition module, the ProbabilityDistribution Vector of gained, the probability distribution are calculated for obtaining the forecast model
Vector is mapped to the probability of each behavior comprising the behavior sequence;
Behavior determining module, for based on the ProbabilityDistribution Vector, probability to be met to the behavior of predetermined predicted condition
It is defined as next time behavior of the user at current time.
As an example, the behavior determining module can include:
Probability comparison module, the probability of each behavior and predetermined probability threshold value are mapped to for the behavior sequence
Magnitude relationship;
The behavior determining module, the probability for being additionally operable to be mapped to any behavior in the behavior sequence are more than the probability
During threshold value, determine that any behavior meets predetermined predicted condition.
In other examples, the forecast model is issued to the user side equipment by server.
As an example, the server is when receiving the application mount request of the user side equipment, by the prediction
The application program packing of model and request installation is issued to the user side equipment and is dealt into the user side equipment.
As an example, the server is after detecting that the forecast model is updated, by under the forecast model after renewal
It is dealt into the user side equipment.
As an example, the server is to before the user side equipment Descend Prediction model, according at least one of following
Forecast model to be issued is chosen from multiple forecast models:
The application type of the user side equipment;
The user characteristics of the user side equipment;
The User action log of the user side equipment.
In other examples, the forecast model is shot and long term memory models.
The function of unit (or module) and the implementation process of effect specifically refer to right in the above method in said apparatus
The implementation process of step is answered, will not be repeated here.
For device embodiment, because it corresponds essentially to embodiment of the method, so related part is real referring to method
Apply the part explanation of example.Device embodiment described above is only schematical, wherein described be used as separating component
The unit or module of explanation can be or may not be physically separate, and the part shown as unit or module can be with
It is or may not be physical location or module, you can with positioned at a place, or multiple network lists can also be distributed to
In member or module.Some or all of module therein can be selected to realize the mesh of this specification scheme according to the actual needs
's.Those of ordinary skill in the art are without creative efforts, you can to understand and implement.
The embodiment of the device of the offer user prompt operation of this specification can be applied on an electronic device.Specifically can be with
Realized by computer chip or entity, or realized by the product with certain function.In a kind of typical realization, electronics is set
Standby is computer, and the concrete form of computer can be personal computer, laptop computer, cell phone, camera phone, intelligence
Can phone, personal digital assistant, media player, navigation equipment, E-mail receiver/send equipment, game console, flat board calculating
Machine, wearable device, internet television, intelligent locomotive, pilotless automobile, intelligent refrigerator, other intelligent home devices or
The combination of any several equipment in these equipment.
Device embodiment can be realized by software, can also be realized by way of hardware or software and hardware combining.With
Exemplified by software is realized, as the device on a logical meaning, being will be non-volatile by the processor of electronic equipment where it
Corresponding computer program instructions read what operation in internal memory was formed in the computer-readable recording mediums such as memory.For hardware view,
As shown in figure 5, a kind of hardware structure diagram of electronic equipment where the device of user prompt operation is provided for this specification, except figure
Outside processor, internal memory, network interface and nonvolatile memory shown in 5, the electronic equipment in embodiment where device
Generally according to the actual functional capability of the electronic equipment, other hardware can also be included, this is repeated no more.
In one embodiment, the memory of electronic equipment can store processor executable program instructions;Processor can
With coupled memory, for reading the programmed instruction of the memory storage, and as response, following operation is performed:Used
The behavior sequence at family, wherein, the behavior sequence includes the behavioural information before current time;Behavior sequence input is deposited
Local forecast model is stored in, user is predicted in the behavior next time at current time;According to prediction result, there is provided be used for
Realize the quick operating mode for the behavior next time predicted.
In other examples, the operation performed by processor may be referred to description related in embodiment of the method above,
It will not go into details for this.
It is above-mentioned that this specification specific embodiment is described.Other embodiments are in the scope of the appended claims
It is interior.In some cases, the action recorded in detail in the claims or step can be come according to different from the order in embodiment
Perform and still can realize desired result.In addition, the process described in the accompanying drawings not necessarily require show it is specific suitable
Sequence or consecutive order could realize desired result.In some embodiments, multitasking and parallel processing be also can
With or be probably favourable.
The preferred embodiment of this specification is the foregoing is only, it is all in this explanation not to limit this specification
Within the spirit and principle of book, any modification, equivalent substitution and improvements done etc., the model of this specification protection should be included in
Within enclosing.